Search your vault
Find files by what they say, not just what they're named. VaultRAG runs
embedding search over AI-generated summaries of every document, so
natural-language queries surface the right content even when you don't
remember the filename. You can still narrow by name when you want to,
and your vault stays a dedicated folder, fully separated from the rest
of your system.
Every major cloud model
Access top-tier models through OpenRouter with no compromises on power.
One key connects you to OpenAI, Anthropic, Google, Meta, and more.
Review each model provider's data policy in OpenRouter and choose the
endpoints that match your requirements.
Or run it all on your computer
Want your files and queries to stay completely private? VaultRAG can run
chat and embedding models directly on your own machine through its built-in
local AI runtime. Download a model in a couple of clicks, then search,
summarize, and chat fully offline with nothing leaving your computer. Switch
between local and cloud whenever you like. Local models are experimental
today and improving fast.
Work with your documents
Pin a file from a search result to keep it handy, drop it into the
model's context to ask questions about it, and dial search depth up or
down for the task at hand. VaultRAG also auto-tunes its retrieval
thresholds as you search, so results get sharper the more you use it.
You still get the desktop workflows people expect, including AirDrop
on Mac and mounted or mapped network drives.
Reads what others can't
Parses Microsoft Office files down to the embedded images inside them:
.docx, .xlsx, .pptx, .pdf, .epub, .msg, .eml, and more.
7-day trial, then one-time purchase — $35
Start free with no credit card. Trial licenses are valid for 7 days
on one device. When you're ready, pay $35 once for the desktop app
and use it on up to two devices.
Updates are included for free. Most users spend just $5-10/month
on AI usage through OpenRouter vs $20/month for ChatGPT Plus or
Claude Pro. Get the same models while actually working with your files,
or run local models and pay nothing for AI usage at all.
New to OpenRouter? See the
OpenRouter setup guide
in the docs.
Tune it your way
VaultRAG summarizes every document on import so search works out of
the box, and you can edit any summary by hand when you want a sharper
description. Customize the system prompt to steer the model's tone,
focus, and behavior to match how you actually work.
Pull in the web when you need it
Ask a question that goes beyond your files and VaultRAG can look it up
on the web in the same chat, then blend those results with what it finds
in your vault. Keep the conversation in one place instead of bouncing
between your documents and a browser.
Always in sync
VaultRAG watches your vault folder and keeps its index current as you
add, rename, move, or delete files, in or out of the app. Whatever is on
disk is what you search, so results never drift out of date and you never
have to rebuild by hand.